改进ORB的轮式机器人视觉导航算法  

Visual navigation algorithm of wheeled robot with improved ORB

作　　者：成怡[1,2] 佟晓宇 朱伟康 CHENG Yi;TONG Xiao-yu;ZHU Wei-kang(School of Electrical Engineering and Automation,Tiangong University,Tianjin 300387,China;Tianjin Key Laboratory of Advanced Electrical Engineering and Energy Technology,Tiangong University,Tianjin 300387,China)

机构地区：[1]天津工业大学电气工程与自动化学院,天津300387 [2]天津工业大学天津市电工电能新技术重点实验室,天津300387

出　　处：《天津工业大学学报》2019年第5期58-63,共6页Journal of Tiangong University

基　　金：天津市自然科学基金资助项目(17JCYBJC18500、17JCYBJC19400);天津市高等学校创新团队培养计划资助项目(TD13-5036)

摘　　要：为实现轮式机器人的自主导航和自主控制,针对传统视觉导航技术中ORB算法的特征点分布均匀程度较低、算法耗时较长的缺点,提出一种改进ORB算法:构建尺度空间金字塔对每层图像进行网格划分,增加空间尺度信息;设置感兴趣区域,采用改进FAST角点提取方法提取特征点;引入非极大值抑制方法,抑制低阈值特征点的输出;基于区域图像特征点分布的方差数值评价待检测图像中特征点分布的均匀程度。实验结果表明:使用改进ORB算法提取到的特征点分布均匀程度较高,匹配效果较好,特征点重叠现象降低,算法耗时较短;当特征点期望数值为3000时,图像中特征点分布的均匀程度提高71.13%,耗时降低比率为69.52%,说明该算法可以更好地应用于轮式机器人执行视觉导航算法。In order to make the wheeled robot realize autonomous navigation and control,aiming at the shortcomings of traditional oriented fast and rotated brief(ORB) algorithm with low uniformity of feature points distribution and long time consuming in visual navigation technology,an improved ORB algorithm was proposed. A scale space pyramid was constructed to mesh each layer of image and improve the spatial scale information;the region of interest was setted and the feature points was extracted with improved FAST corner extraction;a non-maximum suppression method was introduced to suppress the output of low threshold feature points. The uniformity of the feature points distribution in the image to be detected was evaluated based on the variance. The experiment results show that the improved ORB algorithm has higher uniformity of feature points distribution,better matching effects,lower feature point overlap,and shorter consuming time. When the expectation value of feature point is 3 000,the uniformity of the feature point distribution in the image is increased by 71.13%,the time-consuming is decreased by 69.52%. It illustrates that the improved ORB algorithm can be applied to the visual navigation algorithm of wheeled robot with improved performance.

关 键 词：轮式机器人 视觉导航 ORB算法 均匀程度 改进FAST 

分 类 号：TP242.62[自动化与计算机技术—检测技术与自动化装置]